In order to improve the use performance, reduce quality and energy consumption of the winch drum, a method based on response surface method-improved artificial bee colony algorithm was proposed to optimize the drum structure. The finite element was used to simulate the static strength, stiffness and fatigue life of the winch drum; a response surface model was established with the internal material parameters of the drum as the design variable, the stiffness as the constraint, and the minimum fatigue life, the maximum static strength and the quality as the target. Based on the traditional artificial bee colony, the nectar source judgment criterion was improved, which was beneficial to the algorithm to jump out of the local optimum and retain the potential optimum. The idea of mutation was introduced into the algorithm of the search strategy to improve the convergence speed; the improved artificial bee colony algorithm was used for multi-objective solution, and the Pareto frontier solution set of the winch drum was obtained. The results show that the optimized drum mass is reduced by 350.1 kg, the maximum stress is reduced by 20 MPa, and the minimum fatigue life is increased by 5.888 7×105 times, the maximum deformation is reduced by 0.37 mm, so the purpose of optimal design is achieved.